Forty Years of Oil Price Fluctuations: Why the Price of Oil May Still Surprise Us

Author(s): Christiane Baumeister and Lutz Kilian

Source: The Journal of Economic Perspectives , Winter 2016, Vol. 30, No. 1 (Winter
2016), pp. 139-160
Published by: American Economic Association

Stable URL: https://www.jstor.org/stable/43710014

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

is collaborating with JSTOR to digitize, preserve and extend access to The Journal of Economic
Perspectives

This content downloaded from

201.150.98.4 on Thu, 17 Feb 2022 20:58:09 UTC
All use subject to https://about.jstor.org/terms
 Journal of Economic Perspectives - Volume 30, Number 1 - Winter 2016 - Pages 139-160

Forty Years of Oil Price Fluctuations:

Why the Price of Oil May Still Surprise Us

Christiane Baumeister and Lutz Kilian

the emergence of a new regime in the global market for crude oil, in which
It oil the has prices
oil pricesemergence
have beenbeen
largelyhavefree40toyears beeninofresponse
fluctuate a since tolargely new regime
the forces of supplythe free oil to crisis in fluctuate the in global 1973/74, in response market which for to also the crude forces coincided oil, of in supply which with
and demand (Dvir and Rogoff 2010; Alquist, Kilian, and Vigfusson 2013). The
crisis arose when the price of imported oil nearly quadrupled over the course of
a quarter, forcing substantial adjustments in oil-consuming countries. To make
matters worse, some governments in industrialized countries responded by
imposing ceilings on the price of domestically produced crude oil and on the price
of refined oil products such as gasoline, causing gasoline shortages and long lines
at gas stations. In addition, many governments introduced speed limits, banned
automobile traffic on Sundays, or limited retail gasoline purchases (for example,
Ramey and Vine 2011). Hence, pictures of long lines at gas stations and empty
highways have shaped the collective memory of the 1973/74 oil crisis, even though
in reality neither phenomenon was an inevitable consequence of the underlying
rise in the price of crude oil.
Although sharp oil price increases had occurred at irregular intervals throughout
the post-World War II period, as documented in Hamilton (1983, 1985), none of

■ Christiane Baumeister is Assistant Professor of Economics at the University of Notre Dame ,

Notre Dame , Indiana , and Research Affiliate at the Centre of Economic Policy Research ,
London , United Kingdom. Lutz Kilian is Professor of Economics at the University of
Michigan , Ann Arbor ; Michigan , and Research Fellow at the Centre of Economic Policy
Research , London , United Kingdom . Their email addresses are cjsbaumeister@gmail.com
and lkilian@umich.edu.

* For supplementary materials such as appendices, datasets, and author disclosure statements, see th
article page at
http://dx.doi.org/ 10. 1 257/jep.30. 1 . 1 39 doi= 1 0. 1 257/jep.30. 1 . 1 39

This content downloaded from

201.150.98.4 on Thu, 17 Feb 2022 20:58:09 UTC
All use subject to https://about.jstor.org/terms
1 40 Journal of Economic Perspectives

these increases was comparable in magnitude to the increase in the price of oil in
the last quarter of 1973. In fact, prior to 1973, the US price of oil had been regulated
by government agencies, resulting in extended periods of constant oil prices, inter-
rupted only by infrequent adjustments, which tended to coincide with exogenous oil
supply disruptions in the Middle East. This policy resulted in occasional sharp spikes
in the growth rate of the inflation-adjusted price of crude oil.
The US system of oil price regulation came to an end starting in the early
1970s, when the United States no longer had any spare capacity in domestic oil
production to satisfy its growing domestic demand for oil and became increas-
ingly dependent on oil imports from the Middle East, the price of which could
not be regulated domestically (Yergin 1992). When the price of imported crude
oil quadrupled in 1973/74, imposing lower ceilings on the price of domestically
produced crude oil soon proved impractical. The price of oil as measured per
barrel of the West Texas Intermediate (WTI) benchmark - a particular grade
of light and sweet crude oil commonly traded in the United States - rose from
$4.31 per barrel in September 1973 to $10.11 in January 1974. Although the last
vestiges of the regulation of the price of domestic crude oil in the United States
persisted until the early 1980s, for all practical purposes, there was a structural
break in the time series process governing the WTI price of crude oil in early
1974, with the real price of oil fluctuating in response to supply and demand
shocks much like other real industrial commodity prices. It is this modern era of
oil markets that our discussion focuses on.

Figure 1 plots the real price of oil (expressed in March 2015 dollars) star
in January 1974. It shows substantial fluctuations in the real price of oil in r
decades with no obvious long-run trend. The literature has identified a numb
potential determinants of oil price fluctuations, including: 1 ) shocks to global
oil production arising from political events in oil-producing countries, the disc
of new fields, and improvements in the technology of extracting crude oil; 2) shoc
to the demand for crude oil associated with unexpected changes in the global
ness cycle; and 3) shocks to the demand for above-ground oil inventories, refle
shifts in expectations about future shortfalls of supply relative to demand in
global oil market.
In this article, we review the causes of the major oil price fluctuations
1973/74, episode by episode. Although economists have made great strid
recent years in understanding the oil price fluctuations in Figure 1 with the benef
of hindsight, some of the variation in the price of oil over the last 40 years was c
unexpected at the time. We discuss alternative measures of oil price expecta
employed by central banks, by economists, and by households as well as measu
financial market expectations of the price of oil. Although some oil price exp
tions measures can be shown to be systematically more accurate than others, a
price expectations are subject to error. The reason is that even if we understan
determinants of the price of oil, predicting these determinants can be very diffic
in practice. We discuss in the context of concrete examples why it is so difficu
predict the determinants of the price of oil.

This content downloaded from

201.150.98.4 on Thu, 17 Feb 2022 20:58:09 UTC
All use subject to https://about.jstor.org/terms
 Christiane Baumeister and Lutz Kilian 141

Figure 1
Inflation-Adjusted WTI Price of Crude Oil, 1974.1-2015.3

Source: US Energy Information Administration.

Note: The West Texas Intermediate (WTI) oil price series has been deflated with the seasonally adjusted
US consumer price index for all urban consumers.

The gap between the price of oil that was expected and its eventual outcome
represents an oil price "shock." Such surprise changes in the price of oil have been
considered important in modeling macroeconomic outcomes in particular. We
demonstrate how much the timing and magnitude of oil price shocks may change
with the definition of the oil price expectations measure. We make the case that
the oil price expectations measure required for understanding economic decisions
need not be the most accurate measure in a statistical sense, and we illustrate that
the same change in oil prices may be perceived quite differently by households,
policymakers, financial markets, and economists, depending on how they form
expectations. This insight has potentially important implications for understanding
and modeling the transmission of oil price shocks.

Historical Episodes of Major Fluctuations in the Real Price of Oil

The literature on the causes of oil price fluctuations has evolved substantially
since the early 1980s. Initially, all mayor oil price fluctuations were thought to reflect
disruptions of the flow of global oil production associated with exogenous polit-
ical events such as wars and revolutions in OPEC member countries (for example,
Hamilton 2003). 1 Subsequent research has shown that this explanation is only one

1 OPEC refers to Organization of Petroleum Exporting Countries, which was founded in 1960.

This content downloaded from

201.150.98.4 on Thu, 17 Feb 2022 20:58:09 UTC
All use subject to https://about.jstor.org/terms
1 42 Journal of Economic Perspectives

among many, and not as important as originally thought. In fact, most major oil
price fluctuations dating back to 1973 are largely explained by shifts in the demand
for crude oil (for example, Barsky and Kilian 2002, 2004; Kilian 2009a; Kilian and
Murphy 2012, 2014; Bodenstein, Guerrieri, and Kilian 2012; Lippi and Nobili 2012;
Baumeister and Peersman 2013; Kilian and Hicks 2013; Kilian and Lee 2014). 2 By
far the most important determinant of the demand for oil has been shifts in the
flow (or consumption) demand for oil associated with the global business cycle. As
the global economy expands, so does demand for industrial raw materials including
crude oil, putting upward pressure on the price of oil. At times there also have been
important shifts in the demand for stocks (or inventories) of crude oil, reflecting
changes to oil price expectations. Such purchases are not made because the oil
is needed immediately in the production of refined products such as gasoline or
heating oil, but to guard against future shortages in the oil market. Historically,
inventory demand has been high in times of geopolitical tension in the Middle East,
low spare capacity in oil production, and strong expected global economic growth.

The 1973/74 OU Crisis

At first sight, the oil price shock of 1973/74 has the appearance of a negative
shock to the supply of crude oil in that the quantity of crude oil produced fell
in the last quarter of 1973 and the price of oil increased, consistent with a shift
of the supply curve to the left along the demand curve. Indeed, this is the tradi-
tional explanation for this oil price increase advanced by Hamilton (2003). It is
common to refer to the war between Israel and a coalition of Arab countries that

took place between October 6 and 26, 1973, as the cause of this supply shock.
explanation may conjure up images of burning oil fields, but actually there w
fighting in any of the Arab oil-producing countries in 1973 and no oil produ
facilities were destroyed. Instead, this war took place in Israel, Egypt, and S
None of these countries was a major oil producer or a member of OPEC, for
matter. Thus, the disruption of the flow of oil production that took place in
last quarter of 1973 was not a direct effect of the war. Rather, Arab OPEC
tries deliberately cut their oil production by 5 percent starting on October
1973, ten days into the Arab-Israeli War, while raising the posted price of t
oil, followed by the announcement of an additional 25 percent production c
November 5, ten days after the war had ended.
Hamilton (2003) attributes the Arab oil production cuts in October a
November 1973 entirely to the Arab oil embargo against selected Western c
tries, which lasted from October 1973 to March 1974, interpreting this oil em
as an extension of the military conflict by other means rather than an endoge
response to economic conditions. There is, however, an alternative interpret
tion of the same data that does not rely on the war as an explanation. Barsky

2 In related work, Carter, Rausser, and Smith (2011) conclude that similar results hold for comm
prices more generally, noting that commodity price booms over the last four decades have been pre
by unusually high economic growth.

This content downloaded from

201.150.98.4 on Thu, 17 Feb 2022 20:58:09 UTC
All use subject to https://about.jstor.org/terms
 Forty Years of Oil Pńce Fluctuations : Why the Pnce of Oil May Still Surpńse Us 1 43

Kilian (2002) draw attention to the fact that in early 1973 the price of crude oil
received by Middle Eastern oil producers was effectively fixed as a result of the
1971 Tehran/Tripoli agreements between oil companies and governments of oil
producing countries in the Middle East. These five-year agreements set the price
of oil received by the host government for each barrel of oil extracted in exchange
for assurances that the government would allow foreign oil companies to extract as
much oil as they saw fit (Seymour 1980, p. 80). When global demand for oil accel-
erated in 1972-73, reflecting a worldwide economic boom, many Middle Eastern
countries were operating close to capacity already and unable to increase oil output;
whereas others, notably Saudi Arabia and Kuwait, had the spare capacity to increase
their output, and allowed their oil production to be increased, albeit reluctantly.
This reluctance can be attributed to the fact that the posted price agreed upon in
1971 might have been reasonable at the time, but was quickly eroded in real terms
as a result of a depreciating US dollar and rising US inflation. This development
caused increasing Arab opposition to the Tehran/Tripoli agreements that intensi-
fied in March of 1973 and culminated in the repudiation of the agreements on
October 10, 1973, with oil producers deciding to produce less oil at higher prices.
This reaction makes economic sense even in the absence of any monopoly
power by oil producers. Under this interpretation, a substantial fraction of the
observed decline in Arab oil output in late 1973 was simply a reversal of an unusual
increase in Saudi and Kuwaiti oil production that had occurred earlier that year in
fulfillment of the Tehran/Tripoli agreements. Moreover, the decision to reduce oil
production and the objective of raising the oil price was clearly motivated by the
cumulative effects of the dollar devaluation, unanticipated US inflation, and high
demand for oil fueled by strong economic growth, making this oil price increase
endogenous with respect to global macroeconomic conditions.3
There is, of course, no reason to expect the price of oil charged by Arab oil
producers as of January 1974 to be the equilibrium price necessarily. This price
was set on the basis of negotiations among oil producers, not by the market. There
is evidence, however, that the negotiated price was in fact close to the equilibrium
value. A good indication of the shadow price of crude oil is provided by the steady
increase in commonly used indices of non-oil industrial commodity prices between
November 1971 and February 1974. In the absence of contractual constraints,
one would have expected the price of oil to grow at a similar rate in response to
increased global demand. Kilian (2009b), shows that non-oil industrial commodity
prices over this period increased by 75 percent as much as the price of crude oil
(with some individual commodity prices quadrupling, not unlike the price of oil) ,
suggesting that at most 25 percent of the oil price increase of 1973/74 was caused

3 A detailed analysis in Kilian (2008a) shows that the observed changes in the price of oil and in the
quantity of oil produced in the Middle East over the course of the year 1973 is consistent with this inter-
pretation. Notably, only Kuwait and Saudi Arabia reduced their oil output in October 1973 and only to
the extent required to return to normal levels of oil production, suggesting that the war was immaterial
as a motive for the October production cut.

This content downloaded from

201.150.98.4 on Thu, 17 Feb 2022 20:58:09 UTC
All use subject to https://about.jstor.org/terms
1 44 Journal of Economic Perspectives

by exogenous oil supply shocks. This evidence suggests that much of the oil crisis of
1973/74 actually was driven by increased demand for oil rather than reductions in
oil supply. This conclusion is also consistent with the predictions of regressions of
changes in the price of oil on direct measures of exogenous OPEC oil supply shocks
(see Kilian 2008a) . These regressions suggest that it is difficult to explain more than
25 percent of the 1973 oil price increase based on exogenous OPEC supply shocks.

The 1979/80 Oil Crises

The oil crisis of 1973/74 was followed by a second major oil crisis in 1979/80,
when the price of West Texas Intermediate crude oil rose from less than $15 per
barrel in September 1978 to almost $40 in April 1980. As in 1973/74, governments
responded to rising oil prices by rationing gasoline and enforcing price controls,
causing the recurrence of long lines at gas stations. The traditional view, expressed
in Hamilton (2003), has been that this surge in the price of oil was caused by the
reduction in Iranian oil production following the Iranian Revolution. As noted in
Barsky and Kilian (2002), the timing of events casts doubt on this interpretation.
The Iranian Revolution started gradually in late 1978, culminating in the depar-
ture of the Shah of Iran in January 1979 and the arrival of Ayatollah Khomeini
in February 1979. The biggest Iranian production shortfalls occurred in January and
February 1979. Iranian oil production started recovering in March. Given increased
oil production in Saudi Arabia in direct response to the Iranian Revolution, the
shortfall of OPEC oil output in January 1979 was 8 percent relative to September
1978. By April, the shortfall of OPEC output was zero percent. The price of oil did
not increase substantially before May 1979. Even in April 1979, the WTI oil price
was still under $16 per barrel (up about $1 from $14.85 prior to the Iranian Revolu-
tion), yet within a year the WTI price would reach a peak level of $40 per barrel in
April 1980. The same pattern is also found in oil price series not regulated by the
government, such as the US refiners' acquisition cost of crude oil imports. It is not
clear why the effect of an oil supply shock on the price of oil would be delayed for
so long. Thus, the timing of this oil supply shock makes it an unlikely candidate
for explaining the 1979 oil price increase.
As stressed in Kilian and Murphy (2014), this does not mean that the Iranian
Revolution did not matter for the price of oil, but that it mattered because it affected
oil price expectations rather than because it affected the flow of oil production.
Empirical oil market models that allow for both oil demand and oil supply shocks to
affect the price of oil confirm that oil supply shocks played a minor role for the 1979
oil price increase, but suggest that about one-third of the cumulative price increase
was associated with increased inventory demand in anticipation of future oil short-
ages, presumably reflecting geopolitical tensions between the United States and Iran
and between Iran and its neighbors, but also expectations of high future demand for
oil from a booming global economy. This evidence of rising inventory demand
starting in May 1979 is also consistent with anecdotal evidence from oil market partic-
ipants (for example, Yergin 1992). The remaining two-thirds of the cumulative oil
price increase in 1979 are explained by the cumulative effects of flow demand shocks

This content downloaded from

201.150.98.4 on Thu, 17 Feb 2022 20:58:09 UTC
All use subject to https://about.jstor.org/terms
 Christiane Baumeister and Lutz Kilian 1 45

triggered by an unexpectedly strong global economy, not unlike during the first oil
crisis (for example, Kilian 2009a; Kilian and Murphy 2014).

The 1980s and 1990s

Hamilton (2003) also identifies a large exogenous oil supply disruption

ciated with the outbreak of the Iran-Iraq War, which lasted from 1980 unt
In late September 1980, Iraq invaded Iran, causing the destruction of Irani
facilities and disrupting oil exports from both Iran and Iraq. This event was follo
by an increase in the WTI price of oil from $36 per barrel in September to
January 1981, which by all accounts must be attributed to this oil supply disrupt
This episode is instructive because it represents an example of an oil supply
occurring in the absence of major shifts in oil demand. There is little evide
this shock triggering a large price response, consistent with more formal es
from structural oil market models.

The early 1980s saw a systematic decline in the price of oil from its peak i
April 1980. One reason was the shift in global monetary policy regimes toward
more contractionary stance, led by Paul Volcker's decision to raise US interest rate
The resulting global recession lowered the demand for oil and hence the price o
oil. This decline was further amplified by efforts to reduce the use of oil in indu
trialized countries. In addition, declining prospects of future economic growth i
conjunction with higher interest rates made it less attractive to hold stocks of oi
causing a sell-off of the oil inventories accumulated in 1979. Finally, one of the lega-
cies of the first oil crisis had been that numerous non-OPEC countries includin
Mexico, Norway, and the United Kingdom responded to persistently high oil pric
by becoming oil producers themselves or by expanding their existing oil produ
tion. Given the considerable lag between exploration and production, it was only i
the early 1980s, that this supply response to earlier oil price increases became qua
titatively important. OPEC's global market share fell from 53 percent in 1973
43 percent in 1980 and 28 percent in 1985. The increase in non-OPEC oil produc
tion put further downward pressure on the price of oil.
OPEC attempted to counteract the decline in the price of oil in the earl
1980s. Indeed, this is the first time in its history (and the only time) that OPE
took a proactive role in trying to influence the price of oil (also see Almoguera
Douglas, and Herrera 201 1 ) .4 When OPEC agreements to jointly restrict oil produc-
tion in an effort to prop up the price of oil proved ineffective, with many OPE
members cheating on OPEC agreements, as predicted by the economic theory o
cartels (for example, Green and Porter 1984), Saudi Arabia decided to stabili
the price of oil on its own by reducing Saudi oil production. Skeet (1988) offer

4 The literature has not been kind to the view that OPEC since 1973 has acted as a cartel that sets t
price of oil or that controls the price of oil by coordinating oil production among OPEC members
For example, Alhajji and Huettner (2000) stress that OPEC does not fit the theory of cartels. Smith
(2005) finds that there is no conclusive evidence of OPEC acting as a cartel. Cairns and Calfucura (201
conclude that OPEC has never been a functioning cartel. For further discussion also see Almoguera
Douglas, and Herrera (2011) and Colgan (2014).

This content downloaded from

201.150.98.4 on Thu, 17 Feb 2022 20:58:09 UTC
All use subject to https://about.jstor.org/terms
1 46 Journal of Economic Perspectives

a detailed discussion of how these policies were implemented. As Figure 1 shows,

this approach did not succeed. The price of oil continued to fall in the early 1980s,
albeit at a slower pace. The resulting losses in Saudi oil revenue proved so large that
by the end of 1985, Saudi Arabia was forced to reverse its policy of restricting oil
production. The result was a sharp fall in the price of oil in 1986, caused not only
by the resumption of Saudi oil production, but more importantly by a reduction in
inventory demand for oil, given that OPEC had shown itself to be unable to sustain
a higher price of oil (Kilian and Murphy 2014).
Given the abundance of crude oil supplies in the world relative to oil demand,
the ongoing Iran-Iraq War had little effect on the price of oil in the 1980s, notwith-
standing considerable damage to oil shipping in the Persian Gulf with as many as
30 attacks on oil tankers in a given month. It took the invasion of Kuwait in August
of 1990, followed by the Persian Gulf War, which was directed at ejecting Saddam
Hussein from Kuwait, to generate a sharp increase in the price of oil. The disrup-
tion in Iraqi and Kuwaiti oil production associated with this war played an important
role in causing this spike in the price of oil, but an equally important determinant
was higher demand for oil inventories in anticipation of a possible attack on Saudi
oil fields (Kilian and Murphy 2014). Only in late 1990, when the coalition led by the
United States had moved enough troops to Saudi Arabia to forestall an invasion of
Saudi Arabia, these fears subsided and the price of oil fell sharply along with inven-
tory demand. Without this expectational element, it would be difficult to explain
the quick return to lower oil prices in 1991, given that oil production from Kuwait
and Iraq was slow to recover (Kilian 2008a) .
In the late 1990s, the price of oil weakened further. By December 1998, the
WTI price of oil reached an all-time low in recent history of $1 1, when only two years
earlier oil had been trading at $25. This slide was largely associated with reduced
demand for crude oil, arguably caused by the Asian financial crisis of mid-1997,
which in turn was followed by economic crises in other countries including Russia,
Brazil, and Argentina. The recovery in the price of oil starting in 1999 reflected a
combination of factors including higher demand for oil from a recovering global
economy, some cuts in oil production, and increased inventory demand in anticipa-
tion of tightening oil markets (Kilian and Murphy 2014).
This recovery was followed by two major exogenous oil supply disruptions in
late 2002 and early 2003 that in combination rivaled the magnitude of the oil supply
disruptions of the 1970s (Kilian 2008a). One was a sharp drop in Venezuelan oil
production caused by civil unrest in Venezuela; the other was the disruption of
oil production associated with the 2003 Iraq War. The production shortfalls in Iraq
and Venezuela were largely offset by increased oil production elsewhere, however.
Moreover, compared with 1990, there was less concern that this war would affect oil
fields in Saudi Arabia, especially after the US-led ground offensive proved successful,
with missile attacks being the main threat at the beginning of the war. As a result, there
was only a modest shift in inventory demand. Indeed, this oil supply shock episode is
remarkable mainly because the price of oil proved resilient to geopolitical events. The
price of oil only briefly spiked, adding approximately an extra $6 per barrel.

This content downloaded from

201.150.98.4 on Thu, 17 Feb 2022 20:58:09 UTC
All use subject to https://about.jstor.org/terms
 Forty Years of Oil Pńce Fluctuations : Why the Pńce of Oil May Still Surprise Us 147

From the Great Surge of 2003-08 to the Global Financial Crisis

The most remarkable surge in the price of oil since 1979 occurred between
mid-2003 and mid-2008 with the WTI price climbing from $28 to $134 per barrel.
There is widespread agreement that this price surge was not caused by oil supply
disruptions, but by a series of individually small increases in the demand for crude oil
over the course of several years. Kilian (2008b) , Hamilton (2009) , and Kilian and Hicks
(2013), among others, have made the case that these demand shifts were associated
with an unexpected expansion of the global economy and driven by strong additional
demand for oil from emerging Asia in particular. Because oil producers were unable
to satisfy this additional demand, the price of oil had to increase. This view is consistent
with estimates from empirical models of the global oil market, which attribute the bulk
of the cumulative increase in the price of oil to flow demand shocks (Kilian 2009a;
Baumeister and Peersman 2013; Kilian and Murphy 2014). Only in the first months
of 2008 is there any evidence of increased inventory demand (Kilian and Lee 2014) .
An alternative view among some observers has been that this surge in the price
of oil in the physical market is unprecedented and can only be explained as the
result of speculative positions taken by financial traders in the oil futures market.
This literature has been reviewed in depth in Fattouh, Killian, and Mahadeva
(2013). There is no persuasive evidence in support of this financial speculation
hypothesis, which in fact is at odds with estimates of standard economic models of
markets for storable commodities (for example, Alquist and Kilian 2010; Kilian and
Murphy 2014; Knittel and Pindyck forthcoming).
The financial crisis of 2008 illustrates the powerful effects of a sharp drop in the
demand for industrial commodities on the price of these commodities. As orders
for industrial commodities worldwide were sharply curtailed in the second half of
2008 in anticipation of a major global recession, if not depression, the demand for
commodities such as crude oil plummeted, causing a fall in the price of oil from
$134 per barrel in June 2008 to $39 in February 2009. It is noteworthy that shifts
in the demand for industrial commodities such as crude oil may have much larger
amplitude than the corresponding changes in global real GDP because global real
GDP consists to a large extent of consumption, which remained much more stable
during the crisis. When it became clear in 2009 that the collapse of the global finan-
cial system was not imminent, the demand for oil recovered to levels prevailing in
2007, and the price of oil stabilized near $100 per barrel.
There have been a number of smaller demand and supply shocks in the oil
market between 2010 and early 2014. For example, events such as the Libyan
uprising in 2011 were associated with an increase in the price of oil. Kilian and Lee
(2014) estimate that the Libyan crisis caused an oil price increase of somewhere
between $3 and $13 per barrel, depending on the model specification. Likewise,
tensions with Iran in 2012 account for an increase of between $0 and $9 per barrel.
An additional development since 2011 has been a widening of the spread between
the two main benchmark prices for oil, the West Texas Intermediate and Brent
prices, with WTI oil trading at a discount, reflecting a local glut of light sweet crude
oil in the central United States driven by increased US shale oil production (Kilian

This content downloaded from

201.150.98.4 on Thu, 17 Feb 2022 20:58:09 UTC
All use subject to https://about.jstor.org/terms
1 48 Journal of Economic Perspectives

2014a). As a result, the WTI price of crude oil is no longer representative for the
price of oil in global markets, and it has become common to use the price of Brent
crude oil as a proxy for the world price in recent years.
Following a long period of relative price stability, between June 2014 and January
2015 the Brent price of oil fell from $112 to $47 per barrel, providing yet another
example of a sharp decline in the price of oil, not unlike those in 1986 and 2008. In
Baumeister and Kilian (2015a) , we provide the first quantitative analysis of the $49 per
barrel drop in the Brent price between June and December 2014. We conclude that
about $11 of this decline was associated with a decline in global real economic activity
that was predictable as of June 2014 and reflected in other industrial commodity
prices as well. An additional decline in the Brent price of $16 was predictable as of
June 2014 on the basis of shocks to actual and expected oil production that took place
prior to July 2014. These shocks likely reflected the unexpected growth of US shale
oil production but also increased oil production in other countries including Canada
and Russia. The remaining decline of $22 in the Brent price is explained by two shocks
taking place in the second half of 2014. One is a $9 decline explained by a shock to
the storage demand for oil in July 2014; a further $13 decline is explained by an unex-
pected weakening of the global economy in December 2014.

How to Measure Oil Price Expectations

Although economists have made great strides in recent years in understanding

historical oil price fluctuations, as illustrated in the preceding section, some of the
variation in the price of oil over the last 40 years was unexpected at the time. The
extent to which oil price fluctuations are unexpected depends on how expectations
are formed. Below we introduce four alternative measures of oil price expectations
that may be viewed as representative for the oil price expectations of economists,
policymakers, financial market participants, and consumers, respectively.

Economists' Oil Price Expectations

One common approach to constructing oil price expectations is to relate the
price of oil to its own past values as well as past values of other key determinants of
the price of oil suggested by economic theory (for example, Alquist, Kilian, and
Vigfusson 2013). This is the central idea underlying vector autoregression (VAR)
models of the global oil market (Kilian 2008b, 2009a; Baumeister and Peersman
2013). In our empirical analysis, we employ a VAR model specification in the tradi-
tion of Kilian and Murphy (2014) that includes the real price of oil, global crude oil
production, global real economic activity, and changes in global crude oil stocks. We
refer to the implied expectation of the price of oil as the economists' expectation.

Policymakers' Oil Price Expectations

Of course, oil price expectations based on regression models need not be
representative for the views held by financial market participants or by firms and

This content downloaded from

201.150.98.4 on Thu, 17 Feb 2022 20:58:09 UTC
All use subject to https://about.jstor.org/terms
 Christiane Baumeister and Lutz Kilian 149

households in the economy. One possibility is that financial market participant

have more information than can be captured by econometric models. It is equally
possible for financial market participants to ignore, misinterpret, or miss inform
tion captured by model-based oil price predictions, especially if that information
costly to obtain. A natural source of information about the market expectation o
the price of oil is the price of oil futures contracts. Futures contracts are financi
instruments that allow traders to lock in today a price at which to buy a fixed quan-
tity of crude oil at a predetermined date in the future. The most common approac
to inferring the expected price of oil for immediate delivery in the physical market
(also known as the spot price) has been to treat the price of the oil futures contra
of maturity h as the ^-period ahead market expectation of the nominal price of
crude oil. If today's price of a futures contract expiring a year from now is $50, for
example, then the expectation of the spot price of oil one year from now is $50. This
is how the International Monetary Fund forms oil price expectations, and this ha
been common practice at many central banks in the world, as discussed in Alquis
Kilian, and Vigfusson (2013), which is why we refer to this approach as the policy
makers' oil price expectation.

Financial Market Oil Price Expectations

The use of futures prices as measures of market expectations, however, is valid
only if the risk premium - defined as the compensation arbitrageurs receive fo
assuming the price risk faced by hedgers in the oil futures market - is negligible
This assumption is questionable. Hamilton and Wu (2014) document that there is
a large horizon-specific time-varying risk premium in the oil futures market. Th
risk premium varies with the hedging demands of oil producers and refiners and
the willingness of financial investors to take the other side of hedging contracts. The
oil price expectation may be recovered by subtracting the Hamilton-Wu estimate
of the risk premium from the oil futures price for a given horizon. In a comparison
of alternative risk premium estimates, this specification has been shown to produc
the most reliable oil price expectations measure overall (Baumeister and Kilia
2015a). We refer to the latter expectation as the financial market expectation of th
price of oil.
The policymaker's expectation based on the oil futures price and the finan-
cial market expectation proposed in Baumeister and Kilian (2015a) differ no
only quantitatively, but also qualitatively. For expository purposes, our discussion
focuses on expectations of the West Texas Intermediate price of crude oil; analo
gous results could also be constructed for the Brent price. We consider horizons o
up to six months. The sample period starts in January 1988 and extends to the end of
20 1 4. Figure 2A treats the current oil futures price with a maturity of h months as th
expectation for the spot price of oil in h months. In contrast, Figure 2B plots
the corresponding financial market expectation of the price of oil obtained by
subtracting from the current WTI futures price of maturity h the risk premium
estimate for horizon h implied by the term structure model of Hamilton an
Wu (2014).

This content downloaded from

201.150.98.4 on Thu, 17 Feb 2022 20:58:09 UTC
All use subject to https://about.jstor.org/terms
150 Journal of Economic Perspectives

Figure 2
Alternative Expectations Measures Based on WTI Futures Prices

A: Monthly Oil Price Expectations Measure Obtained from the Oil Futures Curve

140-1

120-

1990 1995 2000 2005 2010 2015

B: Monthly Financial Market Oil P

140-.

120 -
..... Expectation I

1990 1995 2000 2005 2010 2015

Notes: In Figure 2A, the expectat

h months is measured by the WT
presence of a risk premium (Bau
spot price for a horizon of h mont
h the estimated risk premium fo
weekly term structure model pr
Kilian (2015a).

This content downloaded from

201.150.98.4 on Thu, 17 Feb 2022 20:58:09 UTC
All use subject to https://about.jstor.org/terms
 Forty Years of Oil Price Fluctuations : Why the Pńce of Oil May Still Surprise Us 151

Figures 2A and 2B illustrate that adjusting the futures price for the risk
premium may matter a lot in measuring oil price expectations. For example, at
the peak of the oil price in mid-2008 and in the subsequent months the term
structure of futures prices in Figure 2A slopes upwards, seemingly implying expec-
tations of rising oil prices, whereas the risk-adjusted futures prices in Figure 2B
indicate expectations of sharply falling oil prices. Only after the spot price of
crude oil fell below about $85 in late 2008, did participants in the futures market
expect the price of oil to recover. Similar patterns can also be found around the
smaller oil price peaks of 2011 and 2012.
Another important difference is that during the long surge in the spot price
between 2003 and early 2008, the futures curve in Figure 2A mostly suggests expec-
tations of falling oil prices, whereas the risk-adjusted futures curve in Figure 2B
often indicates much more plausible expectations of rising oil prices. Only when
the spot price surpassed about $100 per barrel, did the risk-adjusted futures price
become more bearish than the unadjusted futures price. Finally, following the
invasion of Kuwait in 1990, the unadjusted futures curve in Figure 2A suggests
expectations of much more rapidly falling oil prices than the risk-adjusted futures
curve in Figure 2B.

Consumers9 Oil Price Expectations

Yet another approach to measuring oil price expectations is to focus more
directly on the expectations of firms or households. There are no survey data on
the oil price expectations of US households or US manufacturing firms (or of oil
companies, for that matter), but recent research by Anderson, Kellogg, Sallee, and
Curtin (2011) has shown that households in the Michigan Survey of Consumers
typically form expectations about the real (or inflation-adjusted) price of gasoline
according to a simple no-change model such that the nominal gasoline price is
expected to grow at the rate of inflation. Given that the price of gasoline is primarily
determined by the price of crude oil, a reasonable conclusion is that consumers fore-
cast the real and nominal prices of crude oil along much the same lines, allowing us
to proxy consumer expectations about the nominal price of oil based on the current
price of oil and an inflation forecast. Because the inflation expectations data in the
Michigan Survey of Consumers is limited to selected horizons, in practice, we rely
on the fixed coefficient gap model inflation forecast proposed in Faust and Wright
(2013). Given that the inflation component in oil price forecasts is small at short
horizons, as shown in Alquist, Kilian, and Vigfusson (2013), the difference is likely
to be negligible.
The use of such simple prediction rules makes perfect economic sense when
consumers do not have access to more sophisticated oil price forecasts. After
all, consumers cannot be expected to have the time and resources to become
experts in oil price forecasting or to make sense of the range of competing oil price
forecasts produced by experts. Indeed, a good case can be made that in thinking
about and in modeling households' purchase decisions, it is households' own oil
price expectations that matter even if those expectations are not as accurate as some

This content downloaded from

201.150.98.4 on Thu, 17 Feb 2022 20:58:09 UTC
All use subject to https://about.jstor.org/terms
152 Journal of Economic Perspectives

alternative model-based forms of oil price expectations. We refer to this simple rule
of thumb as the consumer oil price expectation.

What Is an Oil Price Shock?

The unanticipated or surprise component of a change in the price of

referred to as an oil price shock. By comparing oil price expectations to
quent outcomes, we may obtain a direct measure of the magnitude of the
shock. Clearly, whether an oil price shock occurred, and how large this s
was, depends on which measure of the oil price expectations we use. This
tion has not received much attention to date. Below we compare: 1) the o
shocks perceived by financial markets, based on the oil price expectations s
Figure 2B; 2) the oil price shocks perceived by policymakers, as shown in Fi
3) oil price shocks perceived by consumers employing a simple rule of thu
predicting oil prices; and 4) oil price shocks as measured by economists em
oil market vector autoregressive (VAR) models. For expository purposes, w
focus on the WTI price of oil.
Figure 3 shows all four measures of oil price shocks of interest for five spe
episodes, starting in 1988.11, namely the invasion of Kuwait in 1990, the Asian
cial crisis of 1997, the 2001/02 oil price slump, the financial crisis in 2008,
2014/15 oil price decline. In each of these episodes, the oil price shocks im
the vector autoregressive (VAR) model are smaller on average than those
by any of the other three expectations measures, but the differences are
pronounced in the last two episodes. For example, following the global fi
crisis, in the last quarter of 2008, the consumer oil price shock was -79 perc
the policymaker oil price shock was -77 percent, compared with only -73 p
based on financial market expectations, suggesting that financial markets w
to anticipate the oil price decline in the fourth quarter somewhat more ac
than consumers and policymakers. Even more striking is that the VAR oil
model predicted a much larger decline in oil prices than financial market
resulting in an oil price shock in the fourth quarter of only -45 percent.
Similarly, neither consumers nor policymakers nor financial markets a
pated the 2014/15 oil price drop, which was reflected in large negative o
shocks of between -7 and -9 percent in the third quarter of 2014, betwee
and -28 percent in the fourth quarter of 2014, and between -22 and -26 p
in the first quarter of 2015. Again, this decline was predicted to a consid
extent by the vector autoregressive (VAR) model, which implies much sm
model-based oil price shocks of -2 percent for 2014.III, -10 percent for 20
and of -18 percent for 2015.1.
Figure 3 illustrates that it can make a difference whether we take the c
er's perspective, the policymaker's perspective, the financial market pers
or the economist's perspective in measuring oil price shocks. The ext
the differences differs by episode. Overall, oil price shocks are largest fr

This content downloaded from

201.150.98.4 on Thu, 17 Feb 2022 20:58:09 UTC
All use subject to https://about.jstor.org/terms
 Christiane Baumeister and Lutz Kilian 153

Figure 3
Quarterly Shocks to Nominal WTI Price of Oil by Episode
(percent)

Notes: Each oil price shock series is constructed by averaging the monthly oil price expectations by
quarter and expressing this average as a percent deviation from the quarterly average of the monthly oil
price outcomes. The policymakers' expectation corresponds to the unadjusted West Texas Intermediate
(WTI) oil futures price. The financial market expectation is constructed by subtracting the Hamilton
and Wu (2014) risk premium estimate from the futures price. The consumer expectation is proxied for
by applying a no-change forecast to the real price of crude oil and adding the expected rate of inflation,
motivated by the results for gasoline price expectations in Anderson, Kellogg, Sallee, and Curtin
(2011). The vector autoregressive model (VAR) expectation is constructed from the reduced-form
representation of the oil market model of Kilian and Murphy (2014) estimated on the full sample. The
model includes an intercept and 24 lags of the real price of oil, the growth rate of global oil production,
a proxy for the change in global crude oil inventories, and a measure of the global business cycle. The
implied predictions are scaled as in Baumeister and Kilian (2014) and converted to dollar terms using
the same expected rate of inflation as that underlying the consumer forecast. The inflation forecasts are
constructed using the fixed-coefficient gap model proposed in Faust and Wright (2014).

consumer's perspective, somewhat smaller from the financial market perspective,

and smallest when viewed through the lens of the vector autoregressive (VAR)
model of the oil market. In fact, policymakers' oil price expectations often are
close to consumers' expectations. This finding is interesting because it suggests
that heterogeneity in oil price expectations and hence in oil price shocks across
economic agents may matter for the transmission of oil price shocks, a possibility
that has not been considered in existing research.
Figure 3 also illustrates another important point, which is that it does not take
large positive oil price shocks to generate a sustained increase in the price of oil.

This content downloaded from

201.150.98.4 on Thu, 17 Feb 2022 20:58:09 UTC
All use subject to https://about.jstor.org/terms
154 Journal of Economic Perspectives

The most persistent surge in the price of oil in modern history occurred between
2003 and mid-2008, as illustrated in Figure 1, yet none of the positive oil price
shocks between 2003.11 and 2008.1 shown in Figure 3 exceeded one standard devia-
tion of the oil price shock series.

Why Is It So Difficult to Anticipate Oil Price Fluctuations?

Although it may seem that economists, policymakers, or financial market partic-

ipants should be able to form accurate expectations about the future price of oil,
if they actually understand the determinants of past oil price fluctuations, Figure 3
illustrates that this is not necessarily the case. The reason is that the price of oil will
only be as predictable as its determinants, even if economic models of the global
oil market are approximately correct. Unless we can foresee the future evolution
of these determinants, surprise changes in the price of oil driven by unexpected
shifts in oil demand or oil supply will be inevitable. This problem arises whether
we formally model the key determinants of the price of oil, as in oil market vector
autoregressive (VAR) models, or whether we rely on an intuitive understanding of
oil markets as financial market participants are prone to.

The Role of Unexpected Demand Shifts Associated with the Global Business Cycle
Economic models of oil markets imply that the price of oil, all else equal,
depends on the state of the global economy. This fact does not make forming oil
price expectations any easier, however, because in practice these expectations can be
only as accurate as our predictions of the evolution of the global business cycle. The
problem is that changes in global real economic activity can be predicted at best at
short horizons and even then only imprecisely. For example, empirical studies have
documented that the predictive accuracy of vector autoregressive (VAR) models of
the global oil market improves during times of persistent and hence predictable
economic expansions or contractions, but is greatly reduced during normal times
(Baumeister and Kilian 2012, 2015b).
The difficulty of forecasting the state of the global economy is illustrated by the
oil market data since 2003. It is now widely accepted that the surge in the price of
oil starting in 2003 was caused primarily by increased demand for crude oil from
emerging Asia and notably China, as these countries industrialized on a large scale,
yet Kilian and Hicks (2013) document that professional forecasters of real GDP
systematically underestimated the extent of Chinese growth time and again for a
period lasting five years. This example shows that not only econometric models, but
also professional forecasters have limited ability to predict the state of the global
economy. One of the difficulties in assessing the prospects for the Chinese economy
even today is that we do not know to what extent Chinese growth after 2003 reflected
a permanent structural transformation of the economy and to what extent it was
fueled by expansionary macroeconomic policies that are not sustainable. Another
challenge is the lack of reliable and timely data for the Chinese economy.

This content downloaded from

201.150.98.4 on Thu, 17 Feb 2022 20:58:09 UTC
All use subject to https://about.jstor.org/terms
 Forty Years of Oil Pńce Fluctuations : Why the Pńce of Oil May Still Surpńse Us 155

The Role of Unexpected Shifts in Global Oil Production

Another key determinant of the price of oil is global oil production. There will
always be oil supply disruptions due to political events in oil-producing countries
that are largely unpredictable. A case in point is the disruption of oil production
caused by the Libyan uprising in 2011 and the subsequent civil war in Libya. Simi-
larly, the task of predicting Iraqi oil production is complicated by the activities of
ISIS and by sectarian violence, neither of which are easily predictable.
It is notjust unexpected disruptions of oil production we need to be concerned
with, however. An even more important task is to gauge the response of global oil
production to surges in the price of oil driven by increased demand for crude
oil. For example, the unprecedented 1973/74 oil price increase was followed by
an equally unprecedented search for new oil fields, which resulted in substantial
increases in non-OPEC oil production in the early 1980s. It was by no means clear
in the 1970s how successful this search for more oil would be or how long it would
take to succeed. Hence, it is not unreasonable to view the oil production increases
of the early 1980s as unexpected oil supply increases or oil supply shocks from the
point of view of oil market participants.
A similar situation arose as the result of the demand-driven oil price surge
between 2003 and mid-2008. Indeed, there was no shortage of skeptics who doubted
the ability of oil producers to satisfy increased demand even in the long run. For
example, proponents of the peak oil hypothesis insisted that global oil production
had permanently peaked by 2007 or that the peak was imminent.5 A case in point
is an IMF study by Benes et al. (2015) that, using data up to 2009, predicted a near
doubling of the price of oil over the coming decade based on the view that geolog-
ical constraints would win out over technological improvements in conserving oil
use and in oil extraction.

There is little doubt that the peak oil hypothesis, taken literally, cannot be righ
because it ignores the fact that at higher oil price levels, more oil production w
be forthcoming, as more expensive extraction technologies become profitable.
example, deep sea oil drilling becomes profitable only at sufficiently high oil pr
Yet, this hypothesis also contains a grain of truth in that as of 2008 no one would h
known for sure whether future oil production would be sufficient to meet dem
at current prices. Even granting that such a supply response did occur following the
1973/74 oil price surge with a delay of five years or more, the obvious questio
2008, as in any similar situation, was whether this time would be different. Nothing
in past experience guaranteed as of 2008 that the oil supply response would b
adequate going forward or that it would occur in a timely manner. For example,
rapid growth of US shale oil production after 2008, which was facilitated in imp
tant part by technological innovations in oil drilling, was a surprise to many analyst

5 The peak oil hypothesis originates with Hubbard (1956) and postulates a bell-shaped curve intende
describe the rate at which crude oil is extracted over time. Once the peak of this curve has been reach
the rate of oil production will decline permanently. This curve may be estimated from past oil pro
tion data. For an economic perspective on the peak oil debate see Holland (2008, 2013).

This content downloaded from

201.150.98.4 on Thu, 17 Feb 2022 20:58:09 UTC
All use subject to https://about.jstor.org/terms
156 Journal of Economic Perspectives

Looking back at this episode, we now know with the benefit of hindsight that
the market for oil worked once again. As in the 1970s, it took about five years from
the peak in the price of oil for the market to generate substantial increases in oil
production on a global scale. This does not mean that increased scarcity will not
become a reality in the long run. In this regard, Hamilton (2013) documented
that historically higher oil production usually reflected the development of oil
fields in new locations, rather than increased efficiency in oil production. The
current US shale oil boom, which is driven by improved technology for horizontal
drilling and fracking, is a counterexample. This boom is unlikely to last forever,
however, even granting that efficiency in shale oil production gains may extend
the length of the boom (Kilian 2014a).
The real question thus is whether demand for oil will diminish when the price
of oil ultimately rises, as firms and consumers substitute alternative fuels for oil
products in the transportation sector, for example. If they do, the peak oil hypoth-
esis will become irrelevant. Indeed, no one today is concerned about the world
running out of coal, yet the "Coal Question" raised byjevons (1866) is eerily remi-
niscent of the peak oil hypothesis of 2007. Jevons stressed that British coal reserves
were finite and would be exhausted by the 1960s if coal consumption were to grow
at the same rate as the population. His predictions proved inaccurate because coal,
which at the time was the primary fuel, was replaced by oil starting in the 1920s and
by other fuels in the 1970s. The question is whether, in the very long run, renewable
energies will do the same to oil products.

The Role of Unexpected Shifts in Inventory Demand

One more difficulty in forming oil price expectations is unanticipated changes
in perceptions about the future scarcity of oil that affect future demand for oil
inventories. Such perceptions may evolve rapidly, for example, in response to
geopolitical or economic crises. Thus, expectations of the price of oil that may have
been perfectly reasonable at the time, may be easily rendered obsolete by unfore-
seen political or economic events. In fact, oil prices may change merely in response
to a shift in uncertainty, reflecting precautionary demand (for example, Adelman
1993; Pindyck 2004; Kilian 2009a; Alquist and Kilian 2010).
Obviously, predicting the exact timing of specific political crises and their
impact on oil markets is next to impossible, even if political analysts have no
trouble identifying geopolitical hotspots. The Arab Spring is a case in point. Like-
wise, the timing of economic crises is difficult to anticipate. An example is the
financial crisis of 2008. Moreover, predicting political crises alone is not suffi-
cient to ensure that oil price expectations are accurate. It is important to keep
in mind that oil inventory demand depends on the shortfall of expected supply
compared with expected demand rather than just one side of the market. Histor-
ical evidence suggests that, in practice, shifts in inventory demand tend to arise
only when geopolitical turmoil coincides with expectations of strong demand for
crude oil and tight oil supplies. Geopolitical tensions alone, in contrast, will not
have an effect on the price of oil as long as oil supplies are plentiful relative to

This content downloaded from

201.150.98.4 on Thu, 17 Feb 2022 20:58:09 UTC
All use subject to https://about.jstor.org/terms
 Christiane Baumeister and Lutz Kilian 157

expected demand (Mabro 1998). For example, as noted earlier, persistent attacks
on oil tankers in the Persian Gulf during the 1980s - at the rate of as many as
30 attacks per month - had no apparent effect on the price of oil. Thus, most
oil price predictions simply ignore the possibility of future political or economic
crises, except to the extent that they are already priced in at the time the predic-
tion is made. Because crises are rare, this strategy usually works, but occasionally
it may result in spectacular predictive failures.

Conclusion

Although our understanding of historical oil price fluctuations has grea

improved, oil prices keep surprising economists, policymakers, consumers,
financial market participants. Our analysis focused on the question of how
this surprise component or shock component of oil price fluctuations is. We
trated that the timing and magnitude of oil price shocks depends on the mea
of oil price expectations. The reason why economists care about oil price shoc
that these shocks affect economic decisions. One channel of transmission is the loss

of discretionary income that is associated with unexpectedly higher oil prices

hence higher gasoline prices). Consumers who are forced to commute to and f
work, for example, often have little choice, but to pay higher gasoline prices, whic
reduces the amount of discretionary income available for other purchases. Ano
channel is that oil price shocks affect expectations about the future path of the pri
of oil. Such expectations enter into net present value calculations of future in
ment projects, the cash flow of which depends on the price of oil. For example
automobile manufacturer's decision of whether to build new production faci
for a sport utility vehicle is directly affected by the price of oil, as is a househ
decision which car to buy. What matters for net present value calculations is not th
magnitude of the oil price surprise in the current period, but the revisio
the expected path of the future price of oil which enters the cash flow. In addi
oil price shocks also affect the cash flow of earlier investment decisions by m
facturing firms. In general, ongoing projects remain profitable as long as the p
exceeds the marginal cost, which depends on the current price of oil. It is ev
possible for higher oil prices to cause ongoing projects to be abandoned (much
a consumer may choose to scrap a gas-guzzling car in response to higher gaso
prices). A more comprehensive review of the transmission of oil price shocks
also takes account of the joint determination of the price of oil and the state of
economy is provided in Barsky and Kilian (2004) and Kilian (2008b, 2014b).
These basic transmission mechanisms have been linked to a wide rang
macroeconomic outcomes including inflation, output, employment, and w
(for example, Kilian 2008, 2014). They also have implications for the deb
about climate change and for environmental policies. Of particular intere
the question of how the effects of oil price shocks vary across industries, pla
and households, how long these effects last, and how they may have changed

This content downloaded from

201.150.98.4 on Thu, 17 Feb 2022 20:58:09 UTC
All use subject to https://about.jstor.org/terms
158 Journal of Economic Perspectives

time. One important insight of the recent literature has been that these ques-
tions cannot be answered without taking account of the underlying causes of
the oil price shock. Our analysis suggests another important direction for future
research. Macroeconomic models of the transmission of oil price shocks to date
have not allowed for heterogeneous oil price expectations across economic agents.
Given the differences in the implied oil price shocks associated with alternative
measures of oil price expectations that we documented, such distinctions may be
of first-order importance for applied work.

■ We thank Ana María Herrera and the editors for helpful discussions.

References

Adelman, Morris A. 1993. The EconomicsAof

Monetary Alternative." NBER Macroeconomics
Petroleum Supply. Cambridge, MA: MIT Press. Annual 2001 , vol. 16, pp. 137-83.
Alhajji, A. F., and David Huettner. 2000. "OPEC Barsky, Robert B., and Lutz Kilian. 2004. "Oil
and other Commodity Cartels: A Comparison." and the Macroeconomy since the 1970s "Journal of
Energy Policy 28(15): 1151-64. Economic Perspectives 18(4): 11 5-34.
Almoguera, Pedro A., Christopher C. Douglas,Baumeister, Christiane, and Gert Peersman.
and Ana María Herrera. 2011. "Testing for the2013. "The Role of Time-Varying Price Elasticities
Cartel in OPEC: Noncooperative Collusion or Just
in Accounting for Volatility Changes in the Crude
Noncooperative?" Oxford Review of Economic Policy
Oil Market." Journal of Applied Econometrics 28(7):
27(1): 144-68. 1087-1109.
Alquist, Ron, and Lutz Kilian. 2010. "What DoBaumeister, Christiane, and Lutz Kilian. 2012.
We Learn from the Price of Crude Oil Futures?" "Real-Time Forecasts of the Real Price of Oů "Journal
Journal of Applied Econometrics 25(4): 539-73. of Business and Economic Statistics 30(2): 326-36.
Alquist, Ron, Lutz Kilian, and Robert J. Baumeister, Christiane, and Lutz Kilian. 2014.
"What Central Bankers Need to Know about Fore-
Vigfusson. 2013. "Forecasting the Price of Oil."
casting Oil Prices." International Economic Review
In Handbook of Economic Forecasting , vol. 2, edited
55(3): 869-89.
by Elliott, Graham, and Allan Timmermann,
427-507. Amsterdam: North-Holland. Baumeister, Christiane, and Lutz Kilian. 2015a.
Anderson, Soren T., Ryan Kellogg, James M. "A General Approach to Recovering Market Expec-
Sallee, and Richard T. Curtin. 2011. "Forecasting tations from Futures Prices with an Application to
Gasoline Prices Using Consumer Surveys." Crude Oil." Unpublished paper, University of Mich-
American Economic Review 101 (3) : 1 10-14. igan. http://www-personal.umich.edu/~lkilian/
Barsky, Robert B., and Lutz Kilian. 2002. "Do We bk4_08 191 5withappendix.pdf.
Really Know that Oil Caused the Great Stagflation? Baumeister, Christiane, and Lutz Kilian. 2015b.

This content downloaded from

201.150.98.4 on Thu, 17 Feb 2022 20:58:09 UTC
All use subject to https://about.jstor.org/terms
 Forty Years of Oil Pńce Fluctuations : Why the Pńce of Oil May Still Surpńse Us 159

"Forecasting the Real Price of Oil in a Changing
World: A Forecast Combination Approach . "Journal
of Business and Economic Statistics 33: 338-51.
Baumeister, Christiane, and Lutz Kilian. Forth-
coming. "Understanding the Decline in the Price
of Oil since June 2014 "Journal of the Association of
Environmental and Resource Economists.
Benes, Jaromir, Marcelle Chauvet, Ondra
Kameník, Michael Kumhof, Douglas Laxton,
"Risk Premia in Crude Oil Futures Prices." Journal
of International Money and Finance 42: 9-37.
Holland, Stephen P. 2008. "Modeling Peak Oil."
Energy Journal 29(2): 61-80.
Holland, Stephen P. 2013. "The Economics
of Peak Oil." In Encyclopedia of Energy , Natural
Resource, and Environmental Economics, vol. 1,
edited by Jason F. Shogren, 146-50. Amsterdam:
Elsevier.

Susanna Mursula, and Jack Selody. 2015. "The Hubbert, M. King. 1956. "Nuclear Energy
Future of Oil: Geology versus Technology." Interna- and the Fossil Fuels." American Petroleum Institute
tional Journal of Forecasting; 31(1): 207-21 . Drilling and Production Practice , Proceedings of Spring
Bodenstein, Martin, Luca Guerrieri, and Lutz Meeting, San Antonio, 7-25.
Kilian. 2012. "Monetary Policy Responses to Oil Jevons, William S. 1886. The Coal Question.
Price Fluctuations." IMF Economic Review 60(4): London: Macmillan and Co.
470-504. Kilian, Lutz. 2008a. "Exogenous Oil Supply
Cairns, Robert D., and Enrique Calfucura. Shocks: How Big Are They and How Much Do
2012. "OPEC: Market Failure or Power Failure?" They Matter for the U.S. Economy?" Review of
Energy Policy 50: 570-80. Economics and Statistics 90(2): 216-40.
Carter, Colin A., Gordon C. Rausser, and Aaron Kilian, Lutz. 2008b. "The Economic Effects of
D. Smith. 2011. "Commodity Booms and Busts."
Energy Price Shocks." Journal of Economic Literature
46(4): 871-909.
Annual Review of Resource Economics 3: 87-1 18.
Colgan, Jeff D. 2014. "The Emperor Has No Kilian, Lutz. 2009a. "Not All Oil Price Shocks
Clothes: The Limits of OPEC in the Global Oil Are Alike: Disentangling Demand and Supply
Shocks in the Crude Oil Market." American
Market." International Organization 68(3): 599-632.
Economic Review 99 (3): 1053-69.
Dvir, Eyal, and Kenneth S. Rogoff. 2010. "Three
Epochs of Oil." Unpublished manuscript, Boston Kilian, Lutz. 2009b. "Comment on 'Causes and
Consequences of the Oil Shock of 2007-08' by
College, August 16. http://scholar.harvard.edu/
files/rogoff/files/ three_epochs_of_oil.pdf. James D. Hamilton." Brookings Papers on Economi
Fattouh, Bassam, Lutz Kilian, and Lavan Activity, 1, Spring, pp. 267-78.
Mahadeva. 2013. "The Role of Speculation in Oil Kilian, Lutz. 2014a. "The Impact of the Shale
Markets: What Have We Learned So Far?" Energy Oil Revolution on U.S. Oil and Gasoline Prices."
Journal 34(3): 7-33. CEPR Discussion Paper 10304.
Faust, Jon, and Jonathan H. Wright 2013. Kilian, Lutz. 2014b. "Oil Price Shocks: Causes
"Forecasting Inflation." In Handbook of Economic Fore- and Consequences." Annual Review of Resource
casting, vol. 2, edited by Elliott, Graham, and Allan Economics 6: 133-54.
Timmermann, 2-56. Amsterdam: North-Holland. Kilian, Lutz, and Bruce Hicks. 2013. "Did Unex-
Green, Edward J., and Robert H. Porter. 1984. pectedly Strong Economic Growth Cause the Oil
"Noncooperative Collusion under Imperfect Price Price Shock of 2003-2008?" Journal of Forecasting
Information." Econometrica 52(1): 87-100. 32(5): 385-94.
Hamilton, James D. 1983. "Oil and the Macro- Kilian, Lutz, and Daniel P. Murphy. 2012.
economy since World War II." Journal of Political "Why Agnostic Sign Restrictions Are Not Enough:
Economy 91(2): 228-48. Understanding the Dynamics of Oil Market VAR
Hamilton, James D. 1985. "Historical Causes of Models." Journal of the European Economic Associa-
Postwar Oil Shocks and Recessions." Energy Journal tion 10(5): 1166-88.
6(1): 97-116. Kilian, Lutz, and Daniel P. Murphy. 2014. "The
Hamilton, James D. 2003. "What Is an Oil Role of Inventories and Speculative Trading in the
Shock?" Journal of Econometrics 113(2): 363-98. Global Market for Crude Oil." Journal of Applied
Hamilton, James D. 2009. "Causes and Conse- Econometrics 29(3): 454-78.
quences of the Oil Shock of 2007-08." Brookings Kilian, Lutz, and Thomas K Lee. 2014. "Quanti-
Papers on Economic Activity, 1, Spring, pp. 215-283. fying the Speculative Component in the Real Price
Hamilton, James D. 2013. "Oil Prices, Exhaust- of Oil: The Role of Global Oil Inventories "Journal
ible Resources and Economic Growth." In of International Money and Finance 42: 71-87.
Handbook on Energy and Climate Change, edited Knittel,
by Christopher R., and Robert S. Pindyck.
Roger Fouquet, 29-57. Cheltenham, UK: Forthcoming.
Edward "The Simple Economics of
Elgar. Commodity Price Speculation." American Economic
Hamilton, James D., andjing Cynthia Wu. 2014. Journal: Macroeconomics.

This content downloaded from

201.150.98.4 on Thu, 17 Feb 2022 20:58:09 UTC
All use subject to https://about.jstor.org/terms
1 60 Journal of Economic Perspectives

Lippi, Francesco, and Andrea Nobili. 2012. "Oil Much Have Things Really Changed?" NBER Macro-
and the Macroeconomy: A Quantitative Structural economics Annual 2010, vol. 25, pp. 333-68.
Analysis." Journal of the European Economic Associa- Seymour, Ian. 1980. OPEC: Instrument of Change.
tion 10(5): 1059-83. London: MacMillan.
Mabro, Robert. 1998. "OPEC Behavior Skeet, Ian. 1988. OPEC: Twenty-Five Years of Prices
1960-1998: A Review of the Literature." Journal
andofPolitics. New York: Cambridge University Press.
Energy Literature 4(1): 3-27. Smith, James. 2005. "Inscrutable OPEC? Behav-
Pindyck, Robert S. 2004. "Volatility and
ioral Tests of the Cartel Hypothesis." Energy Journal
26(1): 51-82.
Commodity Price Dynamics." Journal of Futures
Markets 24(11): 1029-47. Yergin, Daniel. 1992. The Prize: The Epic Quest
Ramey, Valerie A., and Daniel J. Vine. 2011.
for Oil, Money, and Power. New York: Simon and
Schuster.
"Oil, Automobiles, and the U.S. Economy: How

This content downloaded from

201.150.98.4 on Thu, 17 Feb 2022 20:58:09 UTC
All use subject to https://about.jstor.org/terms